Reducing Vs to 1/2 B+

[Ray Ivers]

I've seen this mod mentioned from time to time at various BBS's, so I thought I would post my recent experiences doing it to a Marshall model 2000 amp (schematic link below):

http://www.freeinfosociety.com/elect...iew.php?id=481

There's another revised schematic for this amp with a number of changes made, but this one shows what I found inside my amp (note the BS HV-diode setup; one of each diode pair in my amp was cracked neatly in half ).

I started by disconnecting the input to the choke and connecting this to the PT CT (midway between points A' and D' in the lower-left corner). Here's what happened, none surprisingly:

1) Idle current immediately fell to zero
2) Power output dropped dramatically after rebias - from 240W to 156W (8 ohm tap, 8 ohm load used for all tests)
3) Max-power load line shifted toward the horizontal; switching to the 4-ohm tap brought power back to 196W

Long story short: replacing the original 1K 5W screen resistors with 22 ohm 2W resistors restored power to 235W on the 8 ohm tap. A few other changes were made, to re-range the bias circuit, increase preamp B+, and improve preamp/screen filtering.

At this point, I should mention that my purpose in performing this mod was not to reduce power, but to save the five surviving original KT77's - and the replacement KT77 I was in no hurry to destroy - from the fate that surely awaited them from the (IMO horrible) 690Vp/687Vs original operating conditions, while retaining full-power operation or as close as possible thereto.

I'd have to say that while this mod did indeed reduce power initially, I would not recommend it as a power-reduction technique in the amps (many 100W and higher Marshalls, and surely others as well) in which it would 'drop right in'. The output tubes will be much happier with these operating conditions, though, and should last longer with a greatly reduced chance of failure - and this amp surely sounds every bit as powerful now as it did before, with perhaps a bit more peak power available and a very slight deep-bass reduction from the non-ideal load line. Performing this mod can also increase your options to use different tube types, depending on the voltages involved; even garden-variety 6V6's and EL84's should be usable in a 450Vp/225Vs circuit with no problems (they'll work hard, but they shouldn't fail prematurely).

Ray

[Amp Kat]

Thanks for sharing that Ray and funny you should do that mod because I have to do the exact thing to a Sunn I'm changing over from KT-88's to 6V6's and that exact transformer mod will be used. The Musicman amps use that half power/full power scheme too but most other amps like the newer Marshall's and so many others just lift the Cathodes on two tubes and call it a half power switch when in reality, all us techies know it's only a -3dB cut at the most and just not even close to half power. How would you think this is relative to the back bias mod or even a power scaling mod ? I guess when you get all of the dropping resistor voltages for the preamp just like you want it you can dial in your pregain like you want it. I'm kinda leary about running Pentodes at lower B+ then they are intended for as it seems to suck the life out them a bit in the tonal section but then again it just depends on if you have the luxury of using power tube distortion opposed to preamp drive distortion where I would think the half power mod may not be half bad !

[Ray Ivers]

KB,

Which Sunn is it that you're working on? If it's got a C-L-C filter like the 2000S, you'll need to add another (preferably C-L-C) filter for the preamp/screens, with a 'stacked' first-filter-cap arrangement (like in the schematic I posted) to maintain a balanced current draw from both halves of the PT HV winding. It should work fine, though.

The Music Mans with the voltage-doublers have that cool half-B+ node already built in, too. I actually like the cathode-lift option if it's done right - not for the power reduction, but for the two-tubes-working-real-hard sound.

I'm not a big fan of back bias, as the bias voltage tracks the output stage current (which is always changing when signal is being passed), rather than the PT HV secondary voltage (which changes too, but mainly at high power-output levels, and not nearly to the same extent). You can regulate the back-bias voltage, of course, but as others have recently pointed out, if you don't also regulate the screens, you'll have a steady bias voltage with load-varying Vs and B+, which can make grid-blocking situations even worse. IMO Power Scaling (tracking B+/Vs/bias) is a good idea - I feel it's really the only way to do it well.

I like running high B+ voltages too; lowering the screen voltage allows you to do so without reliability issues, which is kind of nice IMO.

Ray

[Amp Kat]

KB,

Which Sunn is it that you're working on? If it's got a C-L-C filter like the 2000S, you'll need to add another (preferably C-L-C) filter for the preamp/screens, with a 'stacked' first-filter-cap arrangement (like in the schematic I posted) to maintain a balanced current draw from both halves of the PT HV winding. It should work fine, though.

It's a 68 Solarus but both the trannys are fried so I'm using an old Hi-Fi that has the 5 volt recto windings and the B+ is quite to high for 6v6's so I'm using Center Tap to one side. I think it should work fine but we'll see soon.

The Power Scaling is cool but when you get down to under 150 volts it to seems a tad lifeless but you have to figure the thermionic emmissions is reduced so much you wouldn't even think it would conduct but it does. I still have several MOSFET's for the power scaling mods and have built a few. They are excellent for regulating high voltage supplies to a fixed value.

[Ray Ivers]

Dual-B+ Power Supply Schematic

Below is an image I scanned from page 18 of the GEC 'Audio Frequency Amplifier Design' book. The thing I really like about it is that it has the high-B+ circuitry in dotted-line, and the low-B+ section in full-line.



This supply is choke-input, but it could just as easily be capacitor-input; in fact, that's what I'm using, and it works fine. The important thing to keep in mind is that both the high and low-B+ sections must use the same type of filter.

Ray

[Steve Conner]

Hi Ray, all,

The mod sounds good, but I'm wondering if you had considered leaving the plate voltage at 670V while halving the screen voltage? I put a switch on my amp that would choose between 470V screen voltage for 6L6s, KT88s or whatever, and 360V for EL34s. I left the plate voltage at 470V. I once tried 6V6s on the 360V screen voltage setting, and they seemed to hold together. I ought to make an EL84 adapter

[Ray Ivers]

Steve,

I'm wondering if you had considered leaving the plate voltage at 670V while halving the screen voltage?

Do you mean reducing Vs to 1/2 B+? Hey, that's a great idea! Having a switchable hi/lo Vs really does open up your options for using different tube types safely, and the much smaller screen-resistor values that can be used at lower screen voltages seem to reduce output-stage compression slightly (at least to my ears), which can be a good thing in a high-power PA.

Another really cool thing IMO is the greatly reduced PA drive requirement at lower screen voltages. The Marshall 2000 went from -66V bias to -33V, making for an effective 6dB boost in PI/PA gain that made the FX loops more usable.

Ray

[Steve Conner]

Oh, Vs as in Vg2, whoops! It all makes more sense to me now. I originally intended to run all kinds of tubes at 360v screen voltage, but I discovered that the beam tetrode types hardly gave much power, only about 30w before going into grid current and clipping. So I added a switch to bump the screen voltage back up and give the electrons "that last push over the cliff". I always used a 1K screen resistor, so maybe I should reduce it to 470 ohms.

[Ray Ivers]

Steve,

No, you're right - Vg2 is correct - I just assumed Vs had a wider 'recognizability factor' or some such. I think I'll use 'Vscreen' from now on.

The Marshall's power dropped way down at the reduced Vscreen until I changed out the screen resistors to 22R units (basically just stoppers/"fuses"), then it came right back again. If you're making the Vscreen switchable, this is something that probably should be taken into account unless you're looking specifically for a power reduction; also, the preamp should probably be fed with a constant voltage (in the Marshall I used the CLC-filtered screen feed for the preamp too, so I had to reduce the preamp-B+ series dropping resistor value to maintain the preamp B+ at its previous level).

Ray

[Ray Ivers]

One caveat...

I just accidentally shorted the preamp B+ to ground and the 6A mains fuse blew, scaring the crap out of me.

If you're a believer in B+ fuses - as I am - be sure and include another fuse on the CT low-B+ output, as it's a separate HV supply in its own right - and most probably supplying the screen grids, which are known for shorting to anything handy.

Ray